Custom benchmark results are overwritten

Custom benchmark runs, e.g.

> /gcbrun run_experiment.py -a --experiment-config /opt/fuzzbench/service/experiment-config.yaml --experiment-name 2025-01-09-afl3 --fuzzers afl_mb mopt

have their intermediate report results displayed correctly, however once it is finalized, their results are wiped from the report only the default fuzzers are present, see:
https://www.fuzzbench.com/reports/experimental/2025-01-09-afl3/index.html
Needless to say, this is making analyzing assessment results difficult :)

@DonggeLiu @jonathanmetzman